With the Arch Linux valgrind-3.13.0-2 and glibc-2.25-7 packages for i686, no dynamically liked executables seem to work with Valgrind, because glibc's ld-*.so contains instructions that Valgrind can't interpret. $ uname -a Linux thinkpadx60-par32.lan 4.11.12-gnu-1 #1 SMP PREEMPT Mon Jul 24 13:38:40 CEST 2017 i686 GNU/Linux $ valgrind -v /bin/ls ==29025== Memcheck, a memory error detector ==29025== Copyright (C) 2002-2017, and GNU GPL'd, by Julian Seward et al. ==29025== Using Valgrind-3.13.0 and LibVEX; rerun with -h for copyright info ==29025== Command: /bin/ls ==29025== --29025-- Valgrind options: --29025-- -v --29025-- Contents of /proc/version: --29025-- Linux version 4.11.12-gnu-1 (builduser@ebrasca) (gcc version 7.1.1 20170621 (GCC) ) #1 SMP PREEMPT Mon Jul 24 13:38:40 CEST 2017 --29025-- --29025-- Arch and hwcaps: X86, LittleEndian, x86-mmxext-sse1-sse2-sse3 --29025-- Page sizes: currently 4096, max supported 4096 --29025-- Valgrind library directory: /usr/lib/valgrind --29025-- Reading syms from /usr/lib/ld-2.25.so --29025-- Reading syms from /usr/bin/ls --29025-- object doesn't have a symbol table --29025-- Reading syms from /usr/lib/valgrind/memcheck-x86-linux --29025-- object doesn't have a symbol table --29025-- object doesn't have a dynamic symbol table --29025-- Scheduler: using generic scheduler lock implementation. --29025-- Reading suppressions file: /usr/lib/valgrind/default.supp ==29025== embedded gdbserver: reading from /tmp/vgdb-pipe-from-vgdb-to-29025-by-lukeshu-on-??? ==29025== embedded gdbserver: writing to /tmp/vgdb-pipe-to-vgdb-from-29025-by-lukeshu-on-??? ==29025== embedded gdbserver: shared mem /tmp/vgdb-pipe-shared-mem-vgdb-29025-by-lukeshu-on-??? ==29025== ==29025== TO CONTROL THIS PROCESS USING vgdb (which you probably ==29025== don't want to do, unless you know exactly what you're doing, ==29025== or are doing some strange experiment): ==29025== /usr/lib/valgrind/../../bin/vgdb --pid=29025 ...command... ==29025== ==29025== TO DEBUG THIS PROCESS USING GDB: start GDB like this ==29025== /path/to/gdb /bin/ls ==29025== and then give GDB the following command ==29025== target remote | /usr/lib/valgrind/../../bin/vgdb --pid=29025 ==29025== --pid is optional if only one valgrind process is running ==29025== --29025-- REDIR: 0x4019570 (ld-linux.so.2:strlen) redirected to 0x58057083 (???) vex x86->IR: unhandled instruction bytes: 0x67 0xE8 0x6B 0x6A ==29025== valgrind: Unrecognised instruction at address 0x4002e2f. ==29025== at 0x4002E2F: dl_main (in /usr/lib/ld-2.25.so) ==29025== by 0x4016A25: _dl_sysdep_start (in /usr/lib/ld-2.25.so) ==29025== by 0x4001861: _dl_start (in /usr/lib/ld-2.25.so) ==29025== by 0x4000AF6: ??? (in /usr/lib/ld-2.25.so) ==29025== Your program just tried to execute an instruction that Valgrind ==29025== did not recognise. There are two possible reasons for this. ==29025== 1. Your program has a bug and erroneously jumped to a non-code ==29025== location. If you are running Memcheck and you just saw a ==29025== warning about a bad jump, it's probably your program's fault. ==29025== 2. The instruction is legitimate but Valgrind doesn't handle it, ==29025== i.e. it's Valgrind's fault. If you think this is the case or ==29025== you are not sure, please let us know and we'll try to fix it. ==29025== Either way, Valgrind will now raise a SIGILL signal which will ==29025== probably kill your program. ==29025== ==29025== Process terminating with default action of signal 4 (SIGILL): dumping core ==29025== Illegal opcode at address 0x4002E2F ==29025== at 0x4002E2F: dl_main (in /usr/lib/ld-2.25.so) ==29025== by 0x4016A25: _dl_sysdep_start (in /usr/lib/ld-2.25.so) ==29025== by 0x4001861: _dl_start (in /usr/lib/ld-2.25.so) ==29025== by 0x4000AF6: ??? (in /usr/lib/ld-2.25.so) ==29025== ==29025== HEAP SUMMARY: ==29025== in use at exit: 0 bytes in 0 blocks ==29025== total heap usage: 0 allocs, 0 frees, 0 bytes allocated ==29025== ==29025== All heap blocks were freed -- no leaks are possible ==29025== ==29025== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0) ==29025== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0) Illegal instruction (core dumped)
*** This bug has been marked as a duplicate of bug 384230 ***